It is frequently necessary to form a dielectric layer on the surface of a semiconductor to function as an insulating layer or a protective barrier. A common approach to forming such a dielectric involves growing or depositing an oxide layer of some material on the semiconductor surface. Typically, the material is a metal or the semiconducting material itself and the resulting dielectric layer is the corresponding oxide of the material.
In one approach known to the prior art, oxide layers can be formed utilizing the material in the semiconductor itself. These layers are often referred to as "native" oxide layers. In this approach, the oxidation of the semiconductor surface by an oxidizing gas forms the resulting oxide layer. This approach produces favorable results when silicon is involved due to its propensity to form a stable, high quality oxide (SiO.sub.2). However, compound semiconductors composed of two or more of the elements found in Groups IIB, IIIA, IVA, VA and VIA of the Periodic Table are less favorable candidates for such an approach. Compound semiconductors are characterized as having large electron mobilities as well as a range of band gaps that account for their desirable properties as semiconductors. However, compound semiconductors often do not oxidize uniformly or form stable, native oxides.
For example, even though a material such as gallium arsenide is an excellent semiconductor, it is a poor candidate for native oxide formation. The current protocol for inducing native oxide formation on GaAs involves inducing the oxidation of the GaAs in the presence of As.sub.2 O.sub.3 and oxygen. Under such an approach, the resulting oxide layer contains a high number of defects.
Defects occur when the resulting oxide layer is non-uniformly formed. Such defects lead to high surface state densities at the semiconductor/oxide layer interface. These high surface densities trap electrons when a voltage is applied to the semiconductor. This results in concentrations of electrons near the interface which interferes with reproducibility in conductivity. Consequently, undesirable fluctuations in current are produced when a voltage is applied to the semiconductor.
As a result of the foregoing problems in forming native oxide layers on compound semiconductors such as GaAs, other approaches for depositing foreign metal oxides on the surfaces of compound semiconductors have been attempted.
One conventional technique, known in the art as non-reactive sputtering, involves the bombardment of a target material composed of a selected oxide material, such as SiO.sub.2. As the target material is bombarded under a vacuum, the oxide material is physically removed from the target material and is subsequently deposited on the surface of the semiconductor.
Another technique is a plasma-enhanced chemical vapor deposition process wherein vapor phase reactant gases are subjected to radiation, creating an ionized plasma. The ionized reactants subsequently interact to form the desired reaction product which is ultimately deposited on the substrate surface.
A shortcoming of both of the aforementioned approaches is that there is a bombardment of the semiconductor surface by either electrons, ions or the oxide molecules being deposited. This bombardment damages the substrate and results in a high number of defects at the semiconductor/oxide layer interface when attempting to form a metal oxide layer on the surface of compound semiconductors such as GaAs.
Another approach, disclosed in U.S. Pat. No. 4,371,587 to Peters, involves photochemical vapor deposition of the desired oxide material at room temperature. In particular, the Peters reference involves the photochemical generation of neutral oxygen atoms by exposing a non-reactive, oxygen-containing gas to collisions with atoms of mercury vapor that have undergone photochemical excitation. The resulting neutral oxygen atoms are reacted with a compatible vapor phase reactant gas to produce the desired oxide vapor, which subsequently undergoes vapor deposition on the semiconductor surface.
A major shortcoming of the vapor deposition approach previously mentioned is that as the oxide molecules are deposited as a vapor, they have an inherent kinetic energy which is a function of their mass and velocity. When a particle having a kinetic energy is deposited on a surface, kinetic energy is transferred to the receiving surface in the form of heat energy, known as the latent heat of condensation. If this latent heat of condensation is not properly dissipated, it can promote what is known as an exchange reaction at the interface between the oxide layer being deposited and the underlying semiconductor surface.
In an exchange reaction, the latent heat of condensation disrupts the bonds occurring between the atoms which make up individual molecules of the semiconducting substance. This phenomenon typically occurs at the surface where the oxide layer is being deposited. Once these bonds have been broken, the individual atoms present in the semiconductor surface are free to form new bonds. As a result, some of the atoms recombine with atoms in the oxide layer, forming the undesirable defects discussed previously.
As a result, prior art methods which do not adequately address the dissipation of the latent heat of condensation fail to produce a high quality metal oxide layer having a reduced incidence of defects at the dielectric/semiconductor interface.
Furthermore, it is known that inert gases can be retained on the surface of a substrate to buffer the latent heat of condensation which occurs as a result of kinetic energy inherent in particles being deposited on the substrate surface. However, experimental data has suggested that reactive gases, especially those containing oxygen, are undesirable buffers for dissipating the latent heat of condensation because they promote oxidation of the underlying substrate and result in exchange reactions that impair the quality of the dielectric layer. Indeed, the difficulties in using a reactive oxidizing gas are readily apparent in the protocol for creating native oxides of GaAs. Typically, the oxidation of the underlying substrate is difficult to control, despite the fact that oxygen is present predominantly in a gaseous phase rather than being concentrated on the substrate surface as in the present invention.
It is therefore an object of the present invention to provide a new and improved method for depositing an oxide film of a material on a substrate surface by reactive deposition of the material onto the surface in the presence of a solid or liquid layer of an oxidizing gas.
It is also an object of the present invention to provide a high quality metal oxide layer on the surface of a sensitive semiconductor substance, minimizing the incidence of defects at the oxide-semiconductor interface caused by chemical reactions associated with the latent heat of condensation during deposition as well as creating a favorable oxidation environment for reactive deposition.